A Machine Learning-Based Predictive Model for the Management of Incidents in Small and Medium-Sized Enterprises in Peru

• Autores: Cribillero, Luis F., Quispe, Jeyson I., Castañeda, Pedro
• Formato: artículo
• Fecha de Publicación: 2024
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: UPC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorioacademico.upc.edu.pe:10757/676293
• Enlace del recurso: http://hdl.handle.net/10757/676293
• Nivel de acceso: acceso embargado
• Materia: algorithm; Classification

In the context of IT incident management, the prioritization and automation of tickets can be a challenge for companies that lack advanced technologies. However, these difficulties can be overcome today by applying machine learning algorithms and techniques that use historical data to train predictive models, which allows for more efficient and effective IT incident management. The article proposes the implementation of a predictive model that uses machine learning to prioritize IT incidents in these companies. The goal of this proposal is to allow small and medium-sized enterprises to prioritize their incidents automatically, using a model that has been previously trained with a supervised multi-label classification algorithm technique to achieve high accuracy. Experimental results show that the Mean Absolute Error (MAE) is 2.79 and a Mean Squared Error (MSE) of 8.21, using the metrics provided by the scikit-learn library. Additionally, the entropy loss approaches a value of 0, suggesting a precise ability of the model to predict real values. Additionally, an average accuracy level of 93.74% was achieved.